The digital age has changed the way travelers seek destination information through online reviews, creating user-generated content (UGC) data that is invaluable for understanding tourist perceptions. Although previous studies have discussed destination image, most of them used questionnaire methods that tend to be biased, and the existing big data analysis is still limited to cognitive image. This study aims to analyze the cognitive and affective images of Greater Bandung tourist destinations through Google Maps reviews to identify factors that influence tourist perceptions. The research methodology involved collecting 619,113 Google Maps reviews from the top 20 tourist destinations in Greater Bandung, with data preprocessing resulting in 284,498 relevant reviews for analysis. Using the Indonesian language variant of BERT (Bidirectional Encoder Representations from Transformers), the IndoBERT model, our text classification revealed that the 'Atmosphere' dimension dominated cognitive perceptions but showed lower satisfaction compared to 'Natural attractions.' Similarly, in the affective dimensions, 'Exciting' experiences led visitor expressions yet showed lower satisfaction compared to 'Interesting' experiences. The cognitive-affective co-occurrence analysis further revealed strong connections between atmospheric elements and excitement, highlighting how physical attributes drive emotional responses. These data-driven insights enable tourism stakeholders to optimize destination management through targeted improvements in atmospheric quality while leveraging Greater Bandung's natural strengths to enhance its tourism competitiveness.